Chameleon RC5 mode with mem detection enabled and automatic P-States & C-States generation for native power managment

[Mohamed Khairy]

is there any AMD support

[rekursor]

I didn't make any change except the dump of Fsb's values,

So I'm afraid the problem is not about mem speed detection but about cpu FSB detection ... maybe related to speedstep ?

Sorry that I can't help more right now on that one

Hi Rekursor,

 

Just an FYI. The test boot that you provided for me above that got my memory speed read correctly, doesn't seem to be working using the lastest rev (268). It's back to 604MHz instead of 667MHz. I was just curious if the changes with your test boot ever made it into the later versions or not.

 

Thanks.

 

 

EDIT:

So I have a GA-P55-UD3R w/ a core i7 and skill ripjaws rams installed, and I get exactly the same result with asere boot (just tested it right now)

Did you check your smbios.plist file and check that there was no hard-coded serials into it when you tested the asere boot ?

@rednous: what is your dram exact brand ? did you check the dram serial number under windows (i.e. with cpu-z) ?

 

Rekursor, i would not agree with you , cause as far as i remember i had a readable memory serial number with the Asere's boot loader, but this isn't issue for me. I'm reporting this in the name of bug free and functional code

One wish Please, post the newer versions of the compiled Chameleon in the first post. Thanx in advance

[Dabluest]

I didn't make any change except the dump of Fsb's values,

So I'm afraid the problem is not about mem speed detection but about cpu FSB detection ... maybe related to speedstep ?

Sorry that I can't help more right now on that one

 

Gotcha. Thanks for your reply.

[rekursor]

I just added 2 features tonight, for (lga1156) core i5/i7 users:

- Processor interconnection speed is now dynamically set to 4800 GT/s

- default machine is updated to iMac11,1

 

With these new settings, I obtain the perfect mac in the system profiler

 

Now my smbios.plist file only contain the factory serial number as the hardware uuid is also dynamically set

 

(By default, if no smbios.plist exists in /Extra then you get SMserial set to "SOMESRLNMBR", so purists still need smbios.plist to enter a serial number)

 

This new work could easily be expanded (i.e SMbusspeed could be dynamically set for other chipsets, better macpro default smbios_defaults could be added for recent machines ...)

For now, non core i5/i7 users still need to keep their smbios.plist file unmodified, but it's a start!

 

For Developers: Maybe in near future we could remove the need of smbios.plist files by providing the last entry (SMserial) but the algorithm for generating valid ones might be tricky as it should take advantage of the hardware uuid, but still conform to the series number that depends on the machine factory release date ...

 

 

Here's the boot:

boot_10.zip

 

Have fun !

[rednous]

EDIT:

So I have a GA-P55-UD3R w/ a core i7 and skill ripjaws rams installed, and I get exactly the same result with asere boot (just tested it right now)

Did you check your smbios.plist file and check that there was no hard-coded serials into it when you tested the asere boot ?

@rednous: what is your dram exact brand ? did you check the dram serial number under windows (i.e. with cpu-z) ?

 

Hi Rek ,

 

Thanx for answering me again. You're a patient and nice programmer, not yelling at the newbies and non programmers

 

Yep, you're right about the memory modules -- i checked with Asere's last night too. I used to have hardcoded serial numbers in SMBIOS.plist and that was the reason for serial numbers been reported. I have 2 DIMMs of Mushkin and 2 DIMMs of G.Skill installed. Unfortunately i don't have Windows installed -- i hate it , but i'll try to find some disto and test it to see if it'll report the memory serial numbers.

 

P.S. Thanx for the new booter, i'll test it tonight at home and report shortly

[kozlek]

Hey mozodojo,

 

What the ?! "Unsupported CPU: P-States not generated !!!\n"

 

Isn't there enough information available about the processor where the P-states could be looked up in a manually maintained table? Is this something you're working on or are there issues with that approach? Just curious...

 

P-States on new Core ix have only FID. Regarding datasheet it's not "multiplier" but "power state". But if you get P-States from SSDT you'll see - multiplier x bus clock = frequency. If Turbo Mode enabled, P0 always describes turbo mode transition and associated frequency could be any value (must be higher maybe then the max CPU frequency). bits 32-48 from the performance status MSR register is the max P-State control value, and bits 0-15 is the current P-State control value. Couldn't check if it's right because I have no Core ix CPU.

[anibalin]

Neat, now, with boot10, it boots. I can see I get (thanks to wait=yes) that it states something like:

"p-states not generated, unsupported cpu."

(core i5 650)

[ErmaC]

I just added 2 features tonight, for (lga1156) core i5/i7 users:

- Processor interconnection speed is now dynamically set to 4800 GT/s

- default machine is updated to iMac11,1

 

With these new settings, I obtain the perfect mac in the system profiler

 

Now my smbios.plist file only contain the factory serial number as the hardware uuid is also dynamically set

 

(By default, if no smbios.plist exists in /Extra then you get SMserial set to "SOMESRLNMBR", so purists still need smbios.plist to enter a serial number)

 

This new work could easily be expanded (i.e SMbusspeed could be dynamically set for other chipsets, better macpro default smbios_defaults could be added for recent machines ...)

For now, non core i5/i7 users still need to keep their smbios.plist file unmodified, but it's a start!

 

THX rek!

Works also on 1366.

 

rev 173.

 

Systeminfo + my smbios.plist

 

systeminfo ram

 

Fabio

[VCH888]

@mozodojo and others

 

Thanks. Great job.

 

See the picture: EnableC4State vs Gigabyte Dynamic Energy Saver

EP45-UD3R & HD4870, latest FakeSMC.kext w/ FakeSMCRadeon.kext (b9) in PlugIns, Chameleon RC5 r265.

[tonymacx86]

Hey rekursor-

 

Thanks for the changes bro- will test as soon as I get back from vacation Great work for our i3/i5/i7 and x55 systems.

[MacMan@tonymacx86]

I didn't make any change except the dump of Fsb's values,

So I'm afraid the problem is not about mem speed detection but about cpu FSB detection ... maybe related to speedstep ?

Sorry that I can't help more right now on that one

 

 

 

EDIT:

So I have a GA-P55-UD3R w/ a core i7 and skill ripjaws rams installed, and I get exactly the same result with asere boot (just tested it right now)

Did you check your smbios.plist file and check that there was no hard-coded serials into it when you tested the asere boot ?

@rednous: what is your dram exact brand ? did you check the dram serial number under windows (i.e. with cpu-z) ?

Great work, memory detection works perfectly on both of my rigs, size, speed and vendor.

 

CPU detection still needs some work. My overclocked i5-750 gets recognized as a i7 and my i3-530 is recognized as a Core2Duo. Maybe with the new iMacs using i3's now, we can fix that one

[iFIRE]

Hello Rekursor,

 

Thanks for your big work

 

Here work only P-STATES for now, thanks

 

http://img97.imageshack.us/i/28072010050.jpg/

 

http://img185.imageshack.us/i/capturadepantalla201007l.png/

 

GA P55 UD6

i7 860

 

Thanks

[dgobe]

I just added 2 features tonight, for (lga1156) core i5/i7 users:

- Processor interconnection speed is now dynamically set to 4800 GT/s

- default machine is updated to iMac11,1

 

Unfortunately some of the 0x1A processors are 6.4 GT/s, like the 965 and Xeon X55xx.

 

The QPI speed will change with overclocking too when you change the BCLK. Hopefully it should be possible to get the QPI multiplier and bus speed just like CPU-z does instead of returning hard-coded values. I have no idea where to start with that though.

 

I tested your new boot with my memory set to 1333 and 1600 and it detects and displays the speeds dynamically, nice work!

 

@mozodojo

 

Here's what I see for IA32_PERF_STATUS with RT MSR Tool under Windows:

 

 

The processor is idling so it was running at X 12 multiplier, when I load the processor it reads 0x14 or X 20 multiplier.

[anibalin]

Just to add some light.

dp55wb / i5 650.

If I remove smbios it doesnt get detected as imac11.1 but macpro 3.1

 

 

i7's, from what other users posted got detected properly as imac11.1

 

like Osm@rbcn posted, I too got the:

[rednous]

Rek,

 

While booting (Wait=Yes) with Chameleon r. 268 (boot_10.zip you posted earlier) I noticed the following lines repeated twice:

 

Slot 0: Type 19...

.

.

Slot 3: ...

.

CPU is Intel, family 0x6, model 0x17, ext.model 0x1

CPU is Intel, family 0x6, model 0x17, ext.model 0x1

Patched DMI Table

CPU is Intel, family 0x6, model 0x17, ext.model 0x1

CPU is Intel, family 0x6, model 0x17, ext.model 0x1

Patched DMI Table

.

Found SMBIOS System Information Table 1

Customizing SystemID with: ...

.

CPU is Intel, family 0x6, model 0x17, ext.model 0x1

CPU is Intel, family 0x6, model 0x17, ext.model 0x1

Patched DMI Table

.

Read HFS+ file: ...

 

Maybe too many checks or typo?

[rekursor]

please send a screen copy of cpu-z with all your cpu information, and I'll see what I can do ...

Just to add some light.

dp55wb / i5 650.

If I remove smbios it doesnt get detected as imac11.1 but macpro 3.1

...

 

Thanks, should be corrected in my next commit (later tonight)

Rek,

 

While booting (Wait=Yes) with Chameleon r. 268 (boot_10.zip you posted earlier) I noticed the following lines repeated twice:

...

Maybe too many checks or typo?

 

 

Like I said, I handled first the lga1156 based core i5/7, it is highly perfectible, and I know that

Just a litlle bonus for the P55 based mobos users.

(After all, developers can care about their own config too from time to time )

 

Unfortunately some of the 0x1A processors are 6.4 GT/s, like the 965 and Xeon X55xx.

 

The QPI speed will change with overclocking too when you change the BCLK. Hopefully it should be possible to get the QPI multiplier and bus speed just like CPU-z does instead of returning hard-coded values. I have no idea where to start with that though.

 

I tested your new boot with my memory set to 1333 and 1600 and it detects and displays the speeds dynamically, nice work!

 

@mozodojo

 

Here's what I see for IA32_PERF_STATUS with RT MSR Tool under Windows:

 

 

The processor is idling so it was running at X 12 multiplier, when I load the processor it reads 0x14 or X 20 multiplier.

[ErmaC]

P-States on new Core ix have only FID. Regarding datasheet it's not "multiplier" but "power state". But if you get P-States from SSDT you'll see - multiplier x bus clock = frequency. If Turbo Mode enabled, P0 always describes turbo mode transition and associated frequency could be any value (must be higher maybe then the max CPU frequency). bits 32-48 from the performance status MSR register is the max P-State control value, and bits 0-15 is the current P-State control value. Couldn't check if it's right because I have no Core ix CPU.

 

Hello mozodojo.

First of all thanks for the important contribution that you are giving to all.

I wanted to ask, since I am the owner of a LGA1366 i7 CPU on ASUS mobo.

 

How can we help you to identify the problem? Exactly what data you need to fix/complete the generator p-states on that processors?

 

Thanks again

 

Fabio

 

EDIT

I add all the dump collected until now from Asus x58

 

-

-

-

-

- ACPI_960.zip

- ACPI_965.zip

-

-

[anibalin]

Hi Rekursor here is the info you requested.

 

http://localhostr.com/files/9f60cc/i5_1.JPG

http://localhostr.com/files/0cfc4f/i5_2.JPG

http://localhostr.com/files/b9f44a/i5_3.JPG

 

Let me know If I can help.

Respect.

[rekursor]

Thanks,

 

Should be nicer now for your i5 650 (should also fix duplicated string info)

Here:

boot_11.zip

EDIT here (few display bugs fixed):

boot_12.zip

Hi Rekursor here is the info you requested.

 

http://localhostr.com/files/9f60cc/i5_1.JPG

http://localhostr.com/files/0cfc4f/i5_2.JPG

http://localhostr.com/files/b9f44a/i5_3.JPG

 

Let me know If I can help.

Respect.

[klonfuture]

Hi Rekursor,

 

Just an FYI. The test boot that you provided for me above that got my memory speed read correctly, doesn't seem to be working using the lastest rev (268). It's back to 604MHz instead of 667MHz. I was just curious if the changes with your test boot ever made it into the later versions or not.

 

Thanks.

I've got the same 604Mhz mem speed and I think it depends on SpeedStep. Check your CPU FSB, here is mine

[Dabluest]

I've got the same 604Mhz mem speed and I think it depends on SpeedStep. Check your CPU FSB, here is mine

 

Yeah, that was Rekursor's thinking too. I have the FSB already set properly in smbios and my cpu is read correctly, so I went ahead and just added the correct memory speed there as well. Thanks for your info.

 

EDIT: Still trying to figure out why Rev.222 can read it properly while the later revs can't.

[anibalin]

Hi rekursor, tried boot12, but still got (without smbios.plist) core2duo and macpro 3.1:

 

Memory is perfect tough (even serials)

[Trauma!]

With boot 10 and 12 Core i5 gets recognized as core i7 in system profiler. Previous release was working fine.

 

Except that and the fact that Chameleon GUI is really lagging on P55, everything runs fine.

 

Thx

[kozlek]

With boot 10 and 12 Core i5 gets recognized as core i7 in system profiler. Previous release was working fine.

 

Except that and the fact that Chameleon GUI is really lagging on P55, everything runs fine.

 

Thx

 

Some of i5 has model number 0x1E. In windows CPU-z shows Model 0xE, Ext.model 0x1E...

[Trauma!]

Some of i5 has model number 0x1E. In windows CPU-z shows Model 0xE, Ext.model 0x1E...

 

I've just noticed that, and corrected chameleon to fit my needs. We're gonna need an option in boot.plist to force it, isn't ?